How Signx AI works
Signx finds where the fields go and proposes what each one is, and you review the result before anything is sent. This page is for anyone who wants to know how.
How Signx makes sure every field lands in the right place
Ask a language model exactly where something sits and it returns a confident wrong number that no test catches: a signature two centimetres off the line still validates, still renders, still gets signed. So the work is split.
Geometry answers where
Deterministic measurement over the document's text layer. A number that comes from the page itself, not from a guess.
The model answers what and who
The kind of field, who signs it, and whether a row of boxes is a set of alternatives or independent questions.
An anchor joins the two
Every answer points at its own place in the extracted text, so a hallucination fails loudly instead of passing quietly.
You confirm every placement before anything is sent, and low-confidence rows are flagged in the review. If the analysis fails, a scanned file or an empty result, you reach the same placement screen and finish by hand. The degraded path is the normal path minus a head start.
Get your first document signed today
All you need is an email address and a company name.